Optical beam routing apparatus and methods
This invention relates to methods and apparatus for routing light beams in telecommunications devices using holographic techniques, in particular by displaying kinoforms on LCOS (Liquid Crystal on Silicon) devices. Thus we describe optical beam routing apparatus comprising: at least one optical input to receive an input beam; a plurality of optical outputs; a spatial light modulator (SLM) on an optical path between said optical input and said optical outputs; and a driver for said SLM to display a kinoform on said SLM to diffract said input beam into an output beam comprising a plurality of diffraction orders, wherein a routed one of said diffraction orders is directed to at least one selected said optical output; wherein said apparatus is configured to modify a wavefront of said output beam to reduce a coupling of said output beam into said selected optical output; and wherein said kinoform is adapted to compensate for said wavefront modification to compensate for said reduced coupling and thereby to reduce a coupling of other diffracted light from said input beam into others of said optical outputs than said at least one selected optical output.
Latest Cambridge Enterprise Limited Patents:
This invention relates to methods and apparatus for routing light beams in telecommunications devices using holographic techniques. Aspects of the invention relate to displaying kinoforms on LCOS (Liquid Crystal on Silicon) devices.BACKGROUND TO THE INVENTION
Background prior art relating to the use of holographic techniques, in particular kinoforms (phase only holograms), in telecommunications can be found in: U.S. Pat. No. 5,617,227; U.S. Pat. No. 5,416,616; WO03/021341; U.S. Pat. No. 7,457,547; U.S. Pat. No. 7,174,065; GB2,350,961A; GB2,456,170A; US2006/0067611 “Wavelength Selective Reconfigurable Optical Cross-connect”; and WO2007/131649, “Method Of Forming An Image And Image Projection Device”, which describes a method of forming an image comprising providing a device for imparting respective phase-shifts to different regions of an incident wavefront, wherein the phase shifts give rise to an image in a replay field, and causing zero-order light to be focused into a region between the replay field and the device. Further background can be found in “Aberration correction in an adaptive free-space optical interconnect with an error diffusion algorithm”, D. Gil-Leyva, B. Robertson, T. D. Wilkinson, C. J. Henderson, in 1 Jun. 2006/Vol. 45, No. 16/APPLIED OPTICS pp. 3782; in A. A. Neil, M. J. Booth, and T. Wilson, “New modal wavefront sensor: a theoretical analysis”, J. Opt. Soc. Am. A, 17, 1098 (2000); in “Holographic wave front optimisation on an adaptive optical free-space interconnect”, Chapter 7, University of Cambridge PhD dissertation (http://search.lib.cam.ac.uk/?itemid=|depfacaedb|565372), June 2005, by Gil Leyva Zambada D.; and in W. A. Crossland, I. G. Manolis, M. M. Redmond, K. L. Tan, T. D. Wilkinson, M. J. Holmes, T. R. Parker, H. H. Chu, J. Croucher, V. A. Handerek, S. T. Warr, B. Robertson, I. G. Bonas, R. Franklin, C. Stace, H. J. White, R. A. Woolley, and G. Henshall, “Holographic optical switching: the ROSES demonstrator”, J. Lightwave Techn. 18, 1845 (2000).
The grating patterns previously described, used to deflect an optical input beam to output fibres in a LCOS-based switch, have drawbacks arising from insertion loss and crosstalk, due to the limitations in fabricating a perfect SLM and in displaying an ideal phase pattern. This results in significant power being diffracted into unwanted diffraction orders. This is in part due to the non-ideal response of liquid crystal material, and this particular issue is addressed in our co-pending UK patent application no. 1102715.8 filed 16 Feb. 2011.
We now describe techniques which aim to mitigate crosstalk, based on a technique which we will refer to as wavefront encoding.SUMMARY OF THE INVENTION
According to the present invention there is therefore provided optical beam routing apparatus comprising: at least one optical input to receive an input beam; a plurality of optical outputs; a spatial light modulator (SLM) on an optical path between said optical input and said optical outputs; and a driver for said SLM to display a kinoform on said SLM to diffract said input beam into an output beam comprising a plurality of diffraction orders, wherein a routed one of said diffraction orders is directed to at least one selected said optical output; wherein said apparatus is configured to modify a wavefront of said output beam to reduce a coupling of said output beam into said selected optical output; and wherein said kinoform is adapted to compensate for said wavefront modification to compensate for said reduced coupling and thereby to reduce a coupling of other diffracted light from said input beam into others of said optical outputs than said at least one selected optical output.
Broadly speaking, in embodiments of the technique we purposefully introduce an aberration, such as a defocus, into the optical system which distorts the wavefront in a predetermined manner. Then a correction is built into the kinoform displayed on the SLM, which corrects for the designed aberration. The correction process has the effect of reducing the coupling of other diffracted light from the input beam into other optical outputs than the selected output. Thus, for example, the effect of correcting for a selected diffraction order, for example, +1 (or −1) order, is to reduce the coupling of the other diffracted orders into the other optical outputs, in particular in a system in which the outputs are regularly spaced such that the unwanted orders would otherwise overlap with the non-selected outputs. The mathematics of correcting the wavefront distortion at one diffraction order has the effect of increasing the ‘aberration’ for the non-selected orders. Data for the kinoform may be stored in memory in the SLM driver and/or calculated as needed. In embodiments a set of kinoforms may be stored, one to direct light to each optical output (kinoforms may be combined by linear addition).
Techniques for correcting a wavefront using a kinoform are described in Gil-Leyva et al (ibid); a paper by Blanchard and Greenaway demonstrates that combining a grating with a lens results in the diffraction orders being focused at different planes—i.e. optimising focus for, say, the +1 order for a particular output claim defocuses the other orders (P M Blanchard and A H Greenaway, ‘simultaneous multiplane imaging with a distorted diffraction grating’ Applied Optics 38 6692 (1999)). However, as will be described later, it is not necessary for all the optical outputs to be in a common plane.
Although embodiments reduce the coupling of the unwanted diffraction orders into the unselected optical output(s), more generally, embodiments of the techniques we describe effectively locate a ‘hole’ in the noise field around the selected optical output(s). (This may be achieved, for example, using a “ping-pong” type algorithm, as described later). This is useful because for various reasons a real (rather than ideal) SLM displays a non-ideal phase pattern (the SLM is pixilated, the response of the liquid crystal is non-ideal, the SLM mirrors may be imperfect and result in scattered light, and so forth). Thus embodiments of the technique can be employed more generally, to mitigate noise. (We later describe a subset of multicast systems in which preliminary results indicate that cross-talk may not be reduced, but nonetheless there may be advantages in noise reduction).
The skilled person will appreciate that it is to some extent arbitrary as to whether the apparatus or the kinoform is considered as modifying the wavefront, and of the two which is considered as compensating for the modification. Thus, for example, we will describe embodiments of a system in which the apparatus is configured so that a default, non-corrected condition has the optical outputs displaced away from a focal plane of the system, the kinoform introducing lens power to a selected, target diffraction order to correct for this. However in an alternative arrangement the kinoform may, for example, introduce lens power to a selected diffraction order and the apparatus may include a lens, or more particularly a lenslet array, to compensate for this. Likewise, more generally, the kinoform may introduce a phase mask into the system (which may or may not, for example, encode lens power) and the optical system may incorporate a matched filter, for example a diffractive optical element to compensate. Thus the skilled person will appreciate that one of the apparatus and the kinoform is configured to modify the wavefront of the output beam (or beams) and the other of the apparatus or kinoform is configured to compensate for this modification.
In embodiments the apparatus is configured to modify part or all of the wavefront of the output beam by at least a phase of π/2, π or 2π at a wavelength of operation of the apparatus. The full width at half maximum (FWHM) of the selected routed diffraction order may have a dimension or diameter of, say, ω0, in which case the non-selected diffraction orders may have a spot size of at least 2 or 3 times ω0 at a non-selected output or outputs. For a Gaussian beam a corresponding condition may apply to the mode field diameter.
As previously mentioned, in embodiments the apparatus is configured such that absent the wavelength modification compensation the routed diffraction order is defocussed on the selected optical output, the kinoform then including lens power to compensate for this defocus. For example, the apparatus may be configured such that an optical output or output plane of the beam routing apparatus is displaced by at least 0.5%, 0.7%, 1%, 2%, 5%, 10% or 20% of a focal distance of the apparatus (in embodiments, a distance between the optical outputs and a Fourier transform lens of the apparatus). For example, in one embodiment the defocus was approximately 310 μm in 25 mm; in another example the value of s (see later) was 20% of the value of f (see later). Thus in embodiments the apparatus comprises a lens or mirror in an optical path between the SLM and the optical outputs, and the optical outputs are displaced away from a focal plane defined by this lens or mirror, the kinoform compensating for this displacement.
Optionally, in apparatus which is configured by default to be defocussed, a spatial filter may be located at the displaced focal plane to attenuate undiffracted light from the SLM, more particularly to attenuate a zero order region of the output beam.
Additionally or alternatively the apparatus may include a phase mask to modify the wavefront of the output beam, and the kinoform may be configured to compensate for this phase mask, in effect one serving as a matched filter for the other. The phase mask may comprise or consist of a lenslet array (which may be implemented as a diffractive optical element). Additionally or alternatively the phase mask may comprise an axicon array to, by default, displace an output beam into a cone around an optical output (to avoid the optical output).
The skilled person will appreciate that many other alternative phase mask configurations may also be employed: for example a set of phase hologram elements may be employed, one hologram element associated with each optical output. Thus the plurality of optical inputs and outputs may be implemented using a waveguide device that removes the need for a lenslet array. For example (sub)kinoforms can be adapted for a specific wavefront error to ensure that only the +1 or −1 order is coupled efficiently into the selected optical output, the other orders being defocused or aberrated to reduce crosstalk.
The apparatus may have a plurality of optical outputs to provide a multicast or broadcast function. In such an arrangement a different wavefront modification may be applied to light directed to each optical output. The kinoform is configured to direct a selected diffraction order, for example the +1 order, to a plurality of optical outputs, and is adapted to apply a corresponding wavefront compensation (to the selected diffraction order) for each selected optical output. An array of lenses of different focal lengths may be employed to apply the different wavelength modifications to the different optical outputs. Optionally more than one optical input may also be provided.
The skilled person will appreciate that the techniques we describe may be incorporated into apparatus which switches WDM (wavelength division multiplexed) beams.
In embodiments the optical input and outputs may each be provided by a fibre optic (with optical input/output coupling); conveniently a set of optical outputs may be provided by the end of a fibre optic ribbon. In embodiments multiple fibre optic ribbons may be employed to distribute the outputs over a 2D plane, although as described in more detail later, it may be preferable to distribute the optical outputs (fibre optic inputs) over a 3D region of space. Thus, for example, the ends of a set of fibre optic ribbons may be staggered such that each is in a different plane parallel to a plane of the SLM and thus, in effect, the fibre optic ends are distributed over a 3D region (albeit it may be possible to define a tilted 2D plane with respect for the SLM in which the fibre optic ends are located). More generally the ends of a fibre optic array may be arranged to define a set of different planes, to distribute the fibre optic ends in 3D. The skilled person will also appreciate that the optical input and outputs may be displaced transversely with respect to one another, in which case the kinoform may be configured to introduce an angular displacement between the input and output beam or beams. (Such an angular displacement may be implemented by applying a wedge-type phase profile to the SLM either, with a physical wedge or by imposing a modulo 2π phase profile on the kinoform).
In embodiments the SLM driver is configured to provide kinoform data to a drive output for displaying the kinoform on the SLM. The kinoform may be calculated on-the fly, for example using a Gerchberg Saxton-type algorithm but alternatively, because the wavefront modification is predetermined, the compensation may also be predetermined and the appropriate kinoform(s) stored in non-volatile memory. Thus the SLM driver may have a beam selection data input to select one or more optical outputs, and in response to this the data processor may retrieve one or more kinoforms from the non-volatile memory in order to direct the input beam to the selected optical output or outputs and, in addition, apply the appropriate wavefront modification compensation. The non-volatile memory may comprise, for example, Flash memory (which may optionally be programmed, or re-programmed, remotely). However since Flash memory may have a limited life in terms of read/write cycles (˜1 million) it can be preferable to use more reliable memory, such as ferroelectric random access memory (FRAMs) or magnetic random access memory (MRAM). The data processor may be implemented in software, dedicated hardware, or a combination of the two (depending, in part, on whether or not the kinoform is calculated on-the-fly).
A preferred implementation of the apparatus is as an ROADM (Reconfigurable Optical Add Drop Multiplexer) with an optical fibre input and outputs, optionally configured for use in a WDM system, for example by wavelength de-multiplexing, switching, and re-multiplexing. In some preferred embodiments the SLM is an LCOS (liquid crystal on silicon) SLM. Such an SLM is generally reflective but it is possible to make the silicon thin enough for the SLM to be used in a transmissive mode. However the skilled person will appreciate that alternatively the SLM may be, for example, a MEMS (micro-electromechanical system) SLM.
Although some preferred implementations of the apparatus include a Fourier transform lens between the SLM and optical input/outputs, this is not essential. For example the Fourier transform lens may be omitted and a Fresnel lens incorporated into the kinoform displayed on the SLM. As previously mentioned, optionally the input fibre or fibres may be out of plane as compared with the output fibres.
The apparatus may be configured as a wavelength division multiplexed (WDM) optical switch in which the SLM displays a plurality of kinoforms, one for each wavelength, different spatial regions on the SLM display different kinoforms for the respective different wavelengths. It is preferable, though not essential, for the different (sub) kinoforms to be substantially non-overlapping. Then, in embodiments, the apparatus may include first and second line focussing elements, for example cylindrical lenses, in the optical path to/from the SLM (conveniently the SLM is a reflective SLM). The focuses of said first and second focussing elements may then be arranged to be substantially mutually orthogonal to provide wavefront encoding. More particularly an optical multiplexer-demultiplexer may be included prior to the line focussing elements in the path to the SLM (following these elements in the path from the SLM). One of the line focussing elements then providing focussing of the different wavelengths into different positions onto the SLM plane, and the other line focussing element performs the wavefront encoding by introducing an aberration (astigmatism). In this way each different wavelength may be routed to a different selected output and the corresponding (sub)kinoform modified (to include focussing power) to compensate the selected, say +1, order for the wavefront encoding (astigmatism).
Thus in a related aspect the invention provides a wavelength division multiplexed (WDM) optical switch comprising: at least one optical input to receive an input beam; a plurality of optical outputs; a spatial light modulator (SLM) on an optical path between said optical input and said optical outputs; at least one wavelength division multiplexer-demultiplexer on said optical path to and from said SLM; and first and second line focussing elements on said optical path to and from said SLM, wherein each said line focus element is configured to focus light substantially to a line focus, and wherein the line focuses of said first and second focussing elements are substantially mutually orthogonal.
In embodiments the system may also include a driver for the SLM to display a plurality of kinoforms on the SLM, each to diffract a wavelength of the input beam into an output beam comprising a plurality of diffraction orders. For each wavelength a routed order, say +1 or −1, is directed to a selected optical output, and the corresponding kinoform is adapted to compensate for the line focussing of one of the line focussing elements (that is, for the astigmatism). In preferred implementations different spatial regions on the SLM display different (sub)kinoforms.
In a related aspect the invention provides a method of routing an optical beam, the method comprising: receiving at least one input optical beam at a spatial light modulator (SLM); and diffracting said input optical beam by displaying a kinoform on said SLM to direct a routed diffraction order of said diffracted beam to at least one selected optical output of a plurality of optical outputs; wherein the method further comprises: configuring said apparatus to modify a wavefront of said routed diffracted beam to reduce a coupling of said diffracted beam into said selected optical output; and compensating for said wavefront modifying of said routed diffracted beam using said kinoform, to compensate for said reduced coupling such that a coupling of said diffracted light into others of said optical outputs than said at least one selected optical output is reduced.
Thus in embodiments the apparatus is configured to aberrate a wavefront of the routed or target diffracted beam, for example the +1 (or −1) order, and the kinoform corrects for this. In embodiments the configuration comprises defocusing an optical output, and the kinoform encodes a lens to compensate for this. In embodiments the lens is an off-axis Fresnel lens. More generally the apparatus is configured to provide a matched filter for a wavefront modification pattern displayed on the kinoform.
In a multicast system a different matched filter is provided for each optical output and the kinoform displays a wavefront modification pattern to compensate for matched filters corresponding to selected optical outputs. The matched filter may comprise, for example, an array of lenses and/or axicons.
The invention further provides apparatus comprising means for implementing a method as described above.
These and other aspects of the invention will now be further described, by way of example only, with reference to the accompanying Figures:
A primary goal when routing telecom signals with a phase modulating spatial light modulator is to deflect the signal beam (+1 order) from one or more input fibres to specific locations in the output fibre plane with a sufficient efficiency that ensures the switch meets insertion loss specifications, and that the power coupled to other fibres in the replay plane meets desired crosstalk specifications. Here we define crosstalk as the light that we wish to deflect to intended fibre locations that is unintentionally coupled into one or more other fibre locations.
We will describe techniques which purposefully introducing a wavefront error into the system, and correcting for this error by displaying an optimized kinoform on the SLM, to thereby maximize the amount of light coupled into the fiber. Due to the symmetry conditions of a hologram, if we correct for this aberration for the +1 order, all other orders are further aberrated. As a result we can reduce the crosstalk in the switch.
Space-Variant and Space-Invariant Switching Schemes
Optical switches based on LCOS technology may use either optical systems that are based on space-invariant optical configurations, where the input and output arrays are at conjugate Fourier planes, or two planes of space-variant sub-holograms with micro-beams deflected between these planes, with the sub-holograms displaying grating patterns.
To illustrate the first case, consider an LCOS switch 10 based on an input/output fibre array 102, a focusing element 104, such as a lens, and a LCOS SLM 106 arranged in a 2f (or 4f) system, as shown in
Systems constructed from a series of space-invariant optical systems may also be used for WDM (Wavelength Division Multiplexed) systems where a dispersion element, such as a grating, volume hologram, or thin film based filter, is used to de-multiplex and multiplex the beam, and for systems that use reflective as opposed to refractive optics.
Thus in a second type of LCOS based switch 200, the optical system comprises two planes of sub-holograms 206, 210, as shown in
Details of Wavefront Encoding
The use of a kinoform approach, where we display a spatially non-periodic phase pattern, allows deflection to a 3D volume rather than a 2D plane, and in embodiments we use this for wavefront encoding. This approach employs the purposeful introduction of a wavefront error, such as defocus, into the optical system to reduce the amount of crosstalk power that is coupled to the output ports. To correct for this aberration, the pattern displayed on a dynamic hologram is adjusted to ensure optimum coupling of the +1 diffraction order into the desired output port, resulting in defocus of the noise orders.
Referring now to
Using geometric optics we can show that the defocus of the mth order, d(m), is given by
where s is the displacement of the output from the focus, f is the focal length of the Fourier transform lens, and fH is the focal length of the lens on the hologram (kinoform). The condition that the m=+1 order is focused at the output fibre plane for a given defocus value of s is
The position of the +1 beam at the fibre plane, pB, as a function of off-axis lens offset, pL, can be shown to be
Referring now to
The system comprises: A linear single-mode input/output fibre ribbon array 222; a lenslet array 228 having the same pitch as the fibre array and lenslets of substantially identical focal lengths; a collimating lens 230 of focal length f1; a static transmission grating (multiplexer/demultiplexer) 232 that angularly disperses the WDM wavelengths; a cylindrical lens 234 of focal length (f1)/2; and a reflective spatial light modulator 236.
Light enters from the left via the central fibre, and the corresponding lenslet transforms the input mode field radius from 5.2 μm to 50 μm beam waist at the switch input plane, P1. The reason for this mode conversion is to ensure that the beam entering the switch has a divergence to match the SLM beam steering capabilities, the limited dispersion angle of the static de-multiplexing grating, and the requirement to cover a sufficient number of SLM pixels for efficient diffraction. The input beam is collimated by the collimating lens, de-multiplexed by the static diffraction grating into a linear spread of wavelengths which are focused by the cylindrical lens into an array of elliptical beams on the SLM ready to acquire an angular deflection. In one constructed arrangement, beams cover 400 pixels in the y-direction and 5 pixels in the x-direction. Deflection is in the yz plane. The beam then retraces its steps, is multiplexed by the diffraction grating and any angular displacement it has achieved at the SLM is converted into a positional displacement at the lenslet array. The light is then focused down by a lenslet in front of each fibre to maximize coupling efficiency. Thus individual wavelengths to be independently routed as required by displaying a blazed grating (routing to one fibre) or by a hologram to more than one or more fibres using the techniques described in the first patent.
Preferably, in order to ensure that the light launched into each fibre is normal to that fibre, and that the wavelengths focused onto the SLM plane are normal to the SLM for each wavelength, we arrange the system so that the optics are doubly telecentric. This entails that the distance from the plane P1 to the collimating lens=f1, the distance from the collimating lens to the SLM plane=f1, the distance from the collimating lens to the cylindrical lens=(f1)/2, and thus the distance from the cylindrical lens to the SLM=(f1)/2.
The cylindrical lens focuses each wavelength to a beam waist in the x direction, whilst the light remains collimated in the y-direction. The system of
Referring now to
To reduce the crosstalk we introduce a purposeful defocus of the light in the zy plane by placing a second cylindrical lens 252 of focal length f2 that is aligned orthogonally to the original cylindrical lens as shown in
Experimental Validation of Wavefront Encoding
The beam incident of the SLM had a Gaussian beam radius (1/e2 intensity) of 2.4 mm. The incident light was diffracted by phase patterns displayed on the SLM, and a portion of this diffracted light was reflected by the beam splitter, and focused by L4 (f=200 mm) at the replay plane, where a rectangular aperture of dimensions 120 μm×155 μm was used to pass one diffraction order at a time, and the resultant power measured using a large area photodetector. The SLM had a pixel size of 15 μm×15 μm, with 0.5 μm dead-space. The patterns were displayed across 500×500 pixels using 25 discrete phase levels between 0 to 2π.
Initially a set of blazed gratings was defined that deflected the +1 order to one of twelve target positions across the replay plane, F1. The theoretical and measured beam radii were 26.8 μm and 31 μm respectively. These target positions were located at ±200, ±400, ±600, ±800, ±1000, and ±1200 μm from the optical axis. For a blazed grating, the relationship between positioned deflected to, δ, and period, T, is given (for wavelength λ) by
The relationship between target number and physical position is given in Table 1. This relationship takes into account the effect of the relay lenses L2 and L3, which form a de-magnified image of the SLM phase pattern at the intermediate SLM plane 410.
In order to characterize these aperiodic gratings, the power at each test position was measured as the SLM was cycled to deflect the +1 order to all twelve target positions, resulting in a 12×12 power matrix.
A set of wavefront encoded holograms based on spatially non-periodic phase patterns was also calculated to deflect the +1 order to the same transverse positions in plane F2. For this test the pattern was simply defined as an off-axis lens of focal length fH=1.0 meter and offset by a distance pL. This resulted in the replay plane for the +1 order being shifted towards L4 by a distance, s=90 mm.
The terms, yH2, and fH2 represent the effective focal length and lens offset of the de-magnified off-axis lens at the intermediate SLM plane shown on
Thus, if we take into account the relay system lenses, L2 and L3, the de-magnification of the original SLM kinoform pattern increases the defocus. The test aperture was moved to this new plane, and the measurements repeated. The theoretical (calculated using a Gaussian beam model) and measured beam radii are 27.8 μm and 39.5 μm respectively.
Application of Wavefront Encoding in a Fibre Based Switch
To demonstrate the potential of this technique for fibre applications, a model was set up based on the design shown in
We can analyze the system of
To calculate the actual crosstalk power, we should also take into account the optical powers in these orders. Based on the experimental work, the maximum crosstalk of the wavefront encoded pattern is due to the zeroth order (it contains 3.9% of the total power, relatively because the SLM was not anti-reflection coated). This is verified in
Optimization of Kinoform Pattern
The Fourier-transform can be used to design kinoforms for beam-steering switches using an iterative algorithm, such as the Gerchberg-Saxton routine when the replay field is located at the Fourier plane of a lens. In a wavefront encoded system based on the purposeful introduction of defocusing, we no longer have the replay field positioned at the Fourier plane of replay lens, but at some alternative plane longitudinally shifted by a distance s, such that z2=f+s. Thus we should use some other transform to relate the field at the kinoform plane to the replay plane. One such algorithm is the fractional Fourier transform. (An alternative calculation approach is to consider the one or more optical outputs as point sources and to propagate waves back from these to define the desired phase and amplitude at a selected plane, until the kinoform is reached).
Fractional Fourier Transform
The fractional Fourier transform is a well known function that has been used in optics, signal processing, and quantum mechanics. From a pure mathematics perspective, it can be expressed as
The term Aφ is simply a system constant, and when a=1 we have the standard Fourier transform. From inspection we can infer that the form of equation (9, 10) is due to a quadratic phase factor added to a Fourier transform, the same sort of factor a lens imparts on an optical field.
From the Optics Perspective
A Fourier transform directly relates an input field to the spatial frequency components making up that field. There are many texts that describe the fractional Fourier transform as relating the same input field to an intermediate plane that can be described as comprising a combination of spatial and frequency elements [see, for example, H. M. Ozaktas and D. Mendlovic, “Fractional Fourier optics”, J. Opt. Soc. Am. A, 12, pp 743-748 (1995); and L. M. Bernardo, “ABCD matrix formalism of fractional Fourier optics”, Opt. Eng. 35, pp 732-740 (1996)].
Any optical system comprising an input plane, an output plane, and a set of optics in between can be represented using an ABCD matrix (used in ray-tracing and Gaussian beam propagation theory). According to S. A. Collins, “Lens-System Diffraction Integral Written in Terms of Matrix Optics”, J. Opt. Soc. Am, 60, pp 1168-1177 (1970), diffraction through lens systems can be defined in terms of the ABCD matrix that results in an overall expression of the output field with respect to the input field and ABCD matrix coefficients as:
If the ABCD matrix meets certain symmetry conditions (see Collins, ibid), we can rearrange equation (13) to the same form as equation (9-12). Hence we have a fractional Fourier transform. There are two standard configurations, the Lohmann type I and II geometries [A. W. Lohmann, “Image rotation, Wigner rotation, and the fractional Fourier transform”, J. Opt. Soc. Am A, 10, 2181-2186 (1993)]. It is the first (lens positioned halfway between the input and output planes) that we are interested in at the moment as it comes closest to representing wavefront encoding using defocusing. This entails that the distance from the SLM to the focusing lens, and the distance from the focusing lens to the replay plane (+1 order focal plane), both equal f+s, where f is the focal length of the focusing lens, and s is the defocus. If this is the case, then we can express φ as
Let us assume that we have a fractional Fourier-transform system, as shown in
Thus, if s=0, we have a=1 and ξ2=λt and equation (9) simplifies to the standard Fourier Transform relationship for an optical system of the type of
One of the advantages of using a fractional Fourier transform is that it can be expressed in terms of fast Fourier transforms, thereby allowing for rapid calculation and optimization of the replay field as described by Ozaktas et al [H. M. Ozaktas, O. Arikan, M. A. Kutay, and G. Bozdagi, “Digital computation of the fractional Fourier transform”, IEEE Transactions on Signal Processing, 44, 2141-2150 (1996)]. The design of diffractive elements by this fast fractional Fourier transform approach was reported by Zhang et al [Y. Zhang, B. Z. Dong, B. Y Gu, and G. Z. Yang, “Beam shaping in the fractional Fourier transform domain”, J. Opt. Soc. A, 15, 1114-1120 (1998)], and Zalevsky et al [Z. Zalevsky, D. Mendlovic, and R. G. Dorsch, “Gerchberg-Saxton algorithm applied to the fractional Fourier or the Fresnel domain”, Optics Letters 21, 842-844 (1996)]. Their analyses showed that certain sampling criteria should be met to ensure an accurate representation of the replay field. To circumvent this issue one can use the equivalent optical system approach developed by Testorf [M. Testorf, “Design of diffractive optical elements for the fractional Fourier transform domain: phase-space approach”, Appl. Opt. 45, 76-82 (2006)]. This allows calculation of the replay field for any fractional order. In Testorf's analysis, the Lohmann type I system of
where f is the focal length of the lens of
In the paper by L. Bernardo, “ABCD matrix formalism of fractional Fourier optics” (ibid), it is shown how to describe an optical system where the beam illuminating the input plane (the SLM plane) is not planar in terms of a fractional FFT. This is the situation for the wavefront encoded system based on defocusing described above.
Let us consider
where ρH is negative if the beam incident on the hologram is focused to the right of ΣH, and positive appears to come from a virtual focus to the left of ΣH, Equation (18) is derived by applying the thin lens formula to
Thus we can determine the optimum value of z1 such that the system of
This scaled focal length and new value of φ takes into account the nature of the non-planar beam incident on the hologram plane, and with these new parameters we can use the equivalent model representation described previously to calculate the replay field of a quantized SLM in a wavefront encoded switch. With reference to
Note that the above analysis is valid for a transmissive SLM. In the case the reflective SLM of
From the Kinoform Optimization Perspective
As will be shown by example, the fractional FFT fits straightforwardly into “ping pong” algorithms. (Broadly speaking a “ping pong” algorithm comprises initialising a phase distribution for the kinoform, for example randomly or based on an initial target replay field, calculating a replay field of the kinoform, modifying an amplitude distribution of the replay field but retaining the phase distribution, converting this modified replay field to an updated kinoform and then repeating the calculating and modifying to converge on a desired target replay field).
Fourier transforms are fast and therefore well suited to calculating replay field and in optimizing the kinoform phase pattern in standard Fourier plane systems. According to Ozaktas et al [H. M. Ozaktas, O. Arikan, M. A. Kutay, and G. Bozdagi, “Digital computation of the fractional Fourier transform”, IEEE Transactions on Signal Processing, 44, pp 2141-2150 (1996)], it is possible to convert equations (9-12) to a form that uses standard FFTs and IFFTs. There are other algorithms that can be used to calculate the replay field in a wavefront encoded system (direct Fresnel integral for example). However, according to Ozaktas et al, using a Fresnel integral based solution uses O[N2] calculations, whilst using their implementation we use O[N×log(N)] steps. It is faster than other approaches provided that the associated limitations on the optical geometry are acceptable.
Using the equivalent method developed by Testorf we calculate the replay field, El(u,v), using only four steps. Firstly the plane wavefront, which we denote as Ein(x,y) to take into account any amplitude profile, is incident on the SLM (shown in transmission in
Where f1=f2. We write this as E1(u,v)=FrFFT(EH(x,y)). In the actual calculation we use an FFT for step 3, with a spatial sampling corresponding to the N×N pixels of the SLM plane. Thus the fields at all planes are uniformly spatially sampled on an N×N grid, with the sampled u coordinate being given by
where Δ is the pixel size, and n is an integer varying from −N/2 to N/2. The same scaling factor relates v to y. As shown in
In the case of the system of
For further background information on fractional Fourier transforms reference may be made to the following sources: H. M. Ozaktas, “The Fractional Fourier Transform: with Applications in Optics and Signal Processing”, John Wiley & Sons (2001); A. W. Lohmann, “Image rotation, Wigner rotation, and the fractional Fourier transform”, J. Opt. Soc. Am A, 10, pp 2181-2186 (1993); I. Moreno, J. A. Davis, and K. Crabtree, “Fractional Fourier transform optical system with programmable diffractive lenses”, Appl. Opt. 42, pp. 6544-6548 (2003); D. Palima and V. R. Darla, “Holographic projection of arbitrary light patterns with a suppressed zeroth-order beam”, Appl. Opt. 46, pp 4197-4201 (2007); S-C Pei and M-H Yeh, “Two dimensional fractional Fourier transform”, Signal Processing 67, 99-108 (1998); and X. Y. Yang, Q. Tan, X Wei, Y Xiang, Y. Yan, and G. Jin, “Improved fast fractional-Fourier-transform algorithm”, J. Opt. Soc. Am. A, 21, 1677-1681 (2004). Fractional fast Fourier transform code available from the following web sites: www2.cs.kuleuven.be/˜nalag/research/software/FRFT/—for 1D code, and www.ee.bilkent.edy.tr/˜haldun/fracF.m—for 2D code.
Case 1—Optimization of a Kinoform in a Fourier Plane System Using the Gerchberg Saxton Algorithm
For this calculation the SLM comprised a linear array of 400 pixels of pixel size 15 μm, with the SLM illuminated by a collimated Gaussian replay field of beam radius 2 mm at a wavelength of 1550 nm. The replay position is located −0.75 mm from the optical axis, and the phase values were allowed to take any value between 0 and 2π.
Case 2—Optimization of a Kinoform in a Defocused System Using a Modified Gerchberg-Saxton Algorithm
To optimize the replay field using a fractional FFT of order a, written as FrFFT[field, a], we can modify the Gerchberg Saxton “ping-pong” algorithm as follows below (other algorithms, in particular other “ping-pong” algorithms may alternatively be employed).
Here the fractional Fourier transform FrFFT may be implemented using standard FFTs available in off-the-shelf code. We make use of the fact that an inverse FrFFT of a FrFFT[field, a] can be calculated using FrFFT[field, 2-a] [see for example Ozaktas, ibid].
Multi-Casting with FrFFT
Multicasting Using a Lock-and Key Approach
In order to address the above mentioned issue, we can consider interconnecting not to a plane, but to a 3D volume. As fibre ribbons are flat and the fibres are closely packed one approach is to use a flat fibre ribbon and a lenslet array with each lenslet facet having a different focal length, as shown in
Consider a single fiber and lenslet 1 of focal length f1. In an example system the beam from the SLM is focused a distance z1 from the lenslet and has a beam waist of w1. The distance from the fibre to the lenslet is d, and the desired beam waist at the output fibre is wf. The relationship between z1, f1, and w1 and the position of the output waist (z2) and waist size (w2) is a well known function. In the absence of beam clipping (aperture of lenslet<diameter of beam):
For maximum coupling efficiency, z2=d, and w2=wf. The actual percentage of power coupled can be calculated using the mode overlap integral for the fibre.
If we have N output fibres and N output lenslet facets, each with a distinct focal length of fn, then to couple efficiently into nth fibre we should have the incident beam positioned at the desired value of z1(n). This is best explained by referring to FIG. 20. Here we have five incident beams, all focused at an intermediate plane P1 before the lenset array. As each lenslet facet has a different focal length, fn, the beams are focused at different positions with respect to the fibre array (solid rays). As a result, only one beam is coupled efficiently (the top channel). However, if the SLM displays a phase pattern that focuses the light at a different intermediate plane P2 (dashed rays), the bottom channel couples with high efficiency Thus multi-casting to two fibers, n and n′, employs two defocus values, z1(n), and z1(n′) to match the lenslet focal lengths of fn, and fn′. To connect to fibers n and n′ we employ two lens functions of different focal lengths. The underlying idea is that higher orders will not couple efficiently as they are defocused. Thus interconnecting to a 3D volume and using an asymmetric lenslet array should reduce crosstalk when multi-casting (disregarding clipping losses). One can also consider other options, for example an additional diffractive axicon array (an axicon is an element that has a phase profile equivalent to a cone) positioned after the lenslet array that converts the incident Gaussian beam to a circle focus that substantially completely misses the fibre.
Generalized Wavefront Encoding
To illustrate this, refer to the switch 900 shown in
The system 950 of
Other possible wavefront encoding scenarios can also be envisioned, including geometries where either a static or reconfigurable phase distorting mask placed immediately before or after the Fourier transform lens is used to distort the wavefront. The beam steering hologram in such cases should therefore also add a compensating phase profile to the grating or off-axis lens pattern to ensure the final signal beam is Gaussian. Due to symmetry conditions, the other hologram orders will be further aberrated.SUMMARY
Broadly speaking we have described a method of improving performance of WDM switches based on LCOS SLMs by purposefully introducing a wavefront error, such as defocus, into the optical system design. To compensate for this designed aberration we display an optimized kinoform that compensates for the built in error and thereby efficiently focuses the +1 order into the desired output fibre. Higher diffraction orders, on the other hand, are aberrated and fail to couple efficiently into the fibres. This compensation takes the form of a kinoform calculated using a modulo 2π algorithm. By purposefully introducing an aberration into the optical system, such as, but not limited to, defocus, we can significantly reduce the crosstalk power as illustrated in
Further, by purposefully introducing defocus into the optical system we can filter out any remaining zeroth order power by placing a blocking aperture at the zeroth order focal plane, also as illustrated in
We have also described use of interconnecting into a 3D volume instead of onto a 2D plane, where each beam in the switch is focused at a specific point in space as opposed to a specific point on a plane. By using a lenslet array positioned before the fibre array, where each lenslet has a different focal length, we can ensure that only light focused at a specific position in a volume of space is efficiently coupled into a specific fibre. This facilitates multi-casting. Embodiments of this technique can also be implemented by adding aberrating diffractive patterns to the fundamental focusing function of the lenslet array.
Still further, in embodiments there is a reduction of crosstalk due to dynamic effects: By purposefully introducing an aberration into the optical system, such as, but not limited to, defocus, we can significantly reduce the crosstalk power that occurs as holographic patterns are switched between different interconnection patterns.
In embodiments the techniques we have described are particularly to the routing of optical signal beams, in particular in the telecommunications c-band (the 1.5 micron wavelength window), to produce telecommunications devices operating in the near infra-red.
No doubt many effective alternatives will occur to the skilled person. It will be understood that the invention is not limited to the described embodiments and encompasses modifications apparent to those skilled in the art lying within the scope of the claims appended hereto.
1. Optical beam routing apparatus comprising: wherein said kinoform is adapted to compensate for said displacement by said lens or mirror to compensate for said reduced coupling and thereby to reduce a coupling of other diffracted light from said input beam into others of said optical outputs than said at least one selected optical output.
- at least one optical input to receive an input beam;
- a plurality of optical outputs;
- a spatial light modulator (SLM) on an optical path between said optical input and said optical outputs; and
- a driver for said SLM to display a kinoform on said SLM to diffract said input beam into an output beam comprising a plurality of diffraction orders, wherein a routed one of said diffraction orders is directed to at least one selected said optical output;
- wherein said apparatus comprises a lens or mirror arranged between said SLM and said optical outputs in said optical path, wherein said lens or mirror is configured to modify a wavefront of said output beam by displacing said optical outputs away from a focal plane defined by said lens or mirror to reduce a coupling of said output beam into said selected optical output; and
2. Apparatus as claimed in claim 1 wherein said lens or mirror is configured to modify at least part of said wavefront by a phase of at least π/2, π or 2π via said displacing.
3. Apparatus as claimed in claim 1 wherein said other diffracted light comprises others of said diffraction orders than said routed diffraction order.
4. Apparatus as claimed in claim 3 wherein a full width at half maximum (FWHM) spot size of one of said others of said diffraction orders at a said optical output other than said selected optical output is at least twice a FWHM spot size of said routed diffraction order at said selected optical output.
5. Apparatus as claimed in claim 1 wherein said lens or mirror is configured to defocus said routed one of said diffraction orders on said selected optical output, and wherein said kinoform includes lens power to compensate for said defocus.
6. Apparatus as claimed in claim 5 wherein said lens power in said kinoform compensates for said displacement away from said focal plane.
7. Apparatus as claimed in claim 6 further comprising a spatial filter located at said displaced focal plane to attenuate undiffracted light from said SLM.
8. Apparatus as claimed in claim 1, wherein said kinoform is configured to direct light from said input beam to a plurality of said optical outputs, the apparatus further comprising at least one optical element in said optical path between said SLM and said optical outputs configured such that a different said wavefront modification is applied to light directed to each said optical output, and wherein said kinoform is adapted to apply a corresponding said wavefront modification compensation to each selected said optical output.
9. Apparatus as claimed in claim 8 wherein said optical element comprises an array of lenslets of different focal lengths.
10. Apparatus as claimed in claim 1 wherein said driver for said SLM comprises a beam selection data input, a driver output coupled to said SLM, and one or both of non-volatile memory and a data processor coupled between said beam selection data input and said driver output to provide kinoform data for displaying a said kinoform on said SLM, to select one or more said optical outputs responsive to input beam selection data.
11. Apparatus as recited in claim 1 configured as a wavelength division multiplexed (WDM) optical switch, wherein said at least one optical input and said plurality of optical outputs comprise optical fibres, and wherein said SLM is an LCOS (liquid crystal on silicon) SLM.
12. Apparatus as recited in claim 1 configured as a wavelength division multiplexed (WDM) optical switch, wherein said SLM displays a plurality of said kinoforms, one for each wavelength, and wherein different spatial regions on said SLM display different said kinoforms.
13. A method of routing an optical beam, the method comprising:
- receiving at least one input optical beam at a spatial light modulator (SLM); and diffracting said input optical beam by displaying a kinoform on said SLM to direct a routed diffraction order of said diffracted beam to at least one selected optical output of a plurality of optical outputs;
- wherein the method further comprises:
- modifying a wavefront of said routed diffracted beam to reduce a coupling of said diffracted beam into said selected optical output; and
- compensating for said wavefront modifying of said routed diffracted beam using said kinoform, to compensate for said reduced coupling such that a coupling of said diffracted light into others of said optical outputs than said at least one selected optical output is reduced.
14. A method as claimed in claim 13 wherein said modifying comprises defocusing a said optical output and wherein said kinoform is used to encode lens power to compensate for said defocusing.
|5416616||May 16, 1995||Jenkins et al.|
|5617227||April 1, 1997||De Bougrenet De La Tocnaye et al.|
|7174065||February 6, 2007||Lacey et al.|
|7457547||November 25, 2008||Frisken et al.|
|20010050787||December 13, 2001||Crossland|
|20060067611||March 30, 2006||Frisken et al.|
- O'Brien, et al., “A Holographically Routed Optical Crossbar Using a Ferroelectric Liquid-Crystal Over Silicon Spatial Light Modulator”, “Ferroelectrics”, 1996, pp. 79-86, vol. 181, Publisher: Overseas Publishers Association, Published in: MY.
- Bernardo, “ABCD Matrix Formalism of Fractional Fourier Optics”, “Opt. Eng.”, Mar. 1996, pp. 732-740, vol. 35, No. 3, Publisher: Society of Photo-Optical Instrumentation Engineers, Published in: US.
- Gil-Leyva, et al., “Aberration Correction in an Adaptive Free-Space Optical Interconnect With an Error Diffusion Algorithm”, “Applied Optics”, Jun. 1, 2006, pp. 3782-3792, vol. 45, No. 16, Publisher: Optical Society of America, Published in: US.
- Zhang, et al., “Beam Shaping in the Fractional Fourier Transform Domain”, “J. Opt. Soc. Am. A”, 1998, pp. 1114-1120, vol. 15, No. 5, Publisher: Optical Society of America, Published in: US.
- Testorf, “Design of Diffractive Optical Elements for the Fractional Fourier Transform Domain: Phase-Space Approach”, “Applied Optics”, Jan. 1, 2006, pp. 76-82, vol. 45, No. 1, Publisher: Optical Society of America, Published in: US.
- Ozaktas, et al., “Digital Computation of the Fractional Fourier Transform”, “Transactions on Signal Processing”, Sep. 1996, pp. 2141-2150, vol. 44, No. 9, Publisher: IEEE, Published in: US.
- Ozaktas, et al., “Fractional Fourier Optics”, “J. Opt. Soc. Am. A”, Apr. 1995, pp. 743-751, vol. 12, No. 4, Publisher: Optical Society of America, Published in: US.
- Ozaktas, et al., “The Fractional Fourier Transform With Applications in Optics and Signal Processing”, 2001, pp. 418-419, Publisher: John Wiley & Sons, Ltd, Published in: US.
- Moreno, et al., “Fractional Fourier Transform Optical System With Programmable Lenses”, “Applied Optics”, Nov. 10, 2003, pp. 6544-6548, vol. 42, No. 32, Publisher: Optical Society of America, Published in: US.
- Bernardo, et al., “Fractional Fourier Transforms and Imaging”, “J. Opt. Soc. Am. A”, Oct. 1994, pp. 2622-2626, vol. 11, No. 10, Publisher: Optical Society of America, Published in: US.
- Zalevsky, et al., “Gerchberg-Saxton Algorithm Applied in the Fractional Fourier or the Fresnel Domain”, “Optics Letters”, Jun. 15, 1996, pp. 842-844, vol. 21, No. 12, Publisher: Optical Society of America, Published in: US.
- Crossland, et al., “Holographic Optical Switching: The Roses Demonstrator”, “Journal of Lightwave Technology”, Dec. 2000, pp. 1845-1854, vol. 18, No. 12, Publisher: IEEE, Published in: US.
- Palima, et al., “Holographic Projection of Arbitrary Light Patterns With a Supressed Zero-Order Beam”, “Applied Optics”, Jul. 10, 2007, pp. 4197-4201, vol. 46, No. 20, Publisher: Optical Society of America, Published in: US.
- Zambada, “Holographic Wave Front Optimisation on an Adaptive Optical Free-Space Interconnect, Chapter 7, Experimental Optical Interconnect”, “University of Cambridge PhD Dissertation”, Jun. 2005, pp. 99-121.
- Lohmann, “Image Rotation, Wigner Rotation, and the Fractional Fourier Transform”, “J. Opt. Soc. Am. A”, Oct. 1993, pp. 2181-2186, vol. 10, No. 10, Publisher: Optical Society of America, Published in: US.
- Yang, et al., “Improved Fast Fractional-Fourier-Transform Algorithm”, “J. Opt. Soc. Am. A”, Sep. 2004, pp. 1677-1681, vol. 21, No. 9, Publisher: Optical Society of America, Published in: US.
- Collins, “Lens-System Diffraction Integral Written in Terms of Matrix Optics”, “Journal of the Optical Society of America”, Sep. 1970, pp. 1168-1177, vol. 60, No. 9, Publisher: The Optical Society of America, Published in: US.
- Neil, et al., “New Modal Wave-Front Sensor: A Theoretical Analysis”, “J. Opt. Soc. Am. A.”, Jun. 2000, pp. 1098-1107, vol. 17, No. 6, Publisher: Optical Society of America, Published in: US.
- Blanchard, et al., “Simultaneous Multiplane Imaging With a Distorted Grating”, “Applied Optics”, Nov. 10, 1999, pp. 6692-6699, vol. 38, No. 32, Published in: US.
- Pei, et al., “Two Dimensional Discrete Fractional Fourier Transform”, “Signal Processing”, 1998, pp. 99-108, vol. 67, Publisher: Elsevier, Published in: US.
Filed: Mar 7, 2012
Date of Patent: Jun 7, 2016
Patent Publication Number: 20140355985
Assignee: Cambridge Enterprise Limited (Cambridge)
Inventors: Daping Chu (Cambridge), Neil Collings (Cambridge), William Crossland (Harlow), Maura Michelle Redmond (Cambridge), Brian Robertson (Cambridge)
Primary Examiner: Leslie Pascal
Application Number: 14/005,257
International Classification: H04Q 11/00 (20060101); G02F 1/29 (20060101); G03H 1/00 (20060101); G03H 1/22 (20060101); H04J 14/02 (20060101); H04L 12/721 (20130101);